Budget Rig @ 25k

Status
Not open for further replies.
vickybat

vickybat

I am the night...I am...
^^ Amd phenom 2 1055t is no match for core i5 2400. Neither of the hex cores are. Listen mate more cores are good as they facilitate threading but its not that simple. IPC also has to be higher as well as efficiency.

Currently intel trounces amd in both these scenarios. In simple terms, intel's 4 cores are more capable than amd's current 6 cores. They don't deserve a purchase now as they are priced close to sandybridge.

I'm damn sure that SQL server( heck even the mighty oracle server) will fall in front of sandybridge. Try it to believe it.

All the tasks you mentioned are a piece of cake for the i5 2400 and yes its faster than amd's 6 core offerings in all these tasks no matter how heavily threaded they are.

Only in video encoding tasks amd's 6 cores have a small advantage over quadcores ( i would say they are equal) you know apps like handbrake (open source).Overall i5 2400 is the processor to get eyes closed at its price point.

Amd's current scenario is highly skeptical and is not that future proof after bulldozer's dismal performance. On the other hand, intel's ivybridge offereing for 1155 socket is much much more promising and offers a great upgrade path as it will fit all existing h67,h61,p67 and z68 boards.

People who bought the 990fx boards before hand are cursing themselves and are sticking with older phenom 2 cpu's instead of bulldozer. Amd is now heading to launch 2 more chipsets next year i.e 1090fx and 1070fx. This seems to be their latest strategy of launching new chipsets every year's beginning and new cpu's at year end to challenge intel's sucessful tick tock strategy. But all their efforts are falling apart with intel's strong showing.

I don't think i've to say more on this front.

See this as well:

i5 2400 vs phenom 2 1090t anandtech bench
 
Last edited:
OP
krishnandu.sarkar

krishnandu.sarkar

Simply a DIGITian
Staff member
Ya, I've already read all those bechmarks and decided on i5 2400.

Now after reading his comments I just gave a thought.

Ya as far I know, More Cores needed when you are much into gaming or encoding and all, it'll be faster and better. But for tasks it really doesn't mean more cores = better, it's efficiency of each core = better.

Correct me if I'm wrong. ;)
 
Cilus

Cilus

laborare est orare
avichandana20000 said:
want to clear a concept.

The mentioned Gigabyte Board is saying regarding memory--Support for DDR3 1800(O.C.)/1333/1066 MHz memory modules.
This means upto 1333Mhz it will support normally and for 1800 it needs to be overclocked.

Then how will Gskill Ripjaw 4GB 1600 MHz be supported in this board?
Click to expand...

All the current AM3 processor support 1333 MHz ram speed by default. So whatever Ram having higher speed than 1333 MHz, you plug into your motherboard, they will be downclocked to 1333 MHz.

You have to increase the memory multiplier value to 4 in BIOS to enable the XMP-800 profile which makes the Ram to run it at 1600 MHz speed. If your motherboard supports the XMP profile (XMP profiles are present inside a Ram module and contain the information about all the speeds with timing details, the Ram can operate on), then you have to select the XMP-800 profile directly from the mobo.
 
vickybat

vickybat

I am the night...I am...
@ krishnandu.sarkar

Actually encoding is also a task but that task is heavily threaded or in simple terms the process is divided into multiple sub-processes (a.k.a threads) to assist thread level parallelism.
Now each of these threads will go into cpu cores for execution. Lets say there are two cpu's x and y having 4 cores and 6 cores respectively.

You start application z which has 6 threads.

Now processor x will take 4 threads at once while y will take all 6. Lets assume x has a higher IPC than y. Let each thread take 2 seconds for execution per core in cpu x while cpu y takes 6 seconds to process the same thread per core.

Cpu x will take two cycles to execute 6 threads as it has maximum 4 cores. So the total will be 4 secs to execute 6 threads.
Cpu y will execute all 6 threads in parallel and complete the task in 6 seconds in a single cycle.

We see that despite cpu x having a lack of two cores processed the same process two seconds faster than cpu y which had an advantage of two cores.
The only advantage that cpu x had is single threaded performance or simply cpu time required to execute a single thread of m instructions.

I hope you get the point now mate.:smile:
 
Last edited:
MyGeekTips

MyGeekTips

script-kiddie geek
vickybat said:
@ krishnandu.sarkar

Actually encoding is also a task but that task is heavily threaded or in simple terms the process is divided into multiple sub-processes (a.k.a threads) to assist thread level parallelism.
Now each of these threads will go into cpu cores for execution. Lets say there are two cpu's x and y having 4 cores and 6 cores respectively.

You start application z which has 6 threads.

Now processor x will take 4 threads at once while y will take all 6. Lets assume x has a higher IPC than y. Let each thread take 2 seconds for execution per core in cpu x while cpu y takes 6 seconds to process the same thread per core.

Cpu x will take two cycles to execute 6 threads as it has maximum 4 cores. So the total will be 4 secs to execute 6 threads.
Cpu y will execute all 6 threads in parallel and complete the task in 6 seconds in a single cycle.

We see that despite cpu x having a lack of two cores processed the same process two seconds faster than cpu y which had an advantage of two cores.
The only advantage that cpu x had is single threaded performance or simply cpu time required to execute a single thread of m instructions.

I hope you get the point mate now mate.:smile:
Click to expand...

Very fluid explanation even for a 14 year old like me. :)
 
Cilus

Cilus

laborare est orare
vickybat said:
@ krishnandu.sarkar

Actually encoding is also a task but that task is heavily threaded or in simple terms the process is divided into multiple sub-processes (a.k.a threads) to assist thread level parallelism.
Now each of these threads will go into cpu cores for execution. Lets say there are two cpu's x and y having 4 cores and 6 cores respectively.

You start application z which has 6 threads.

Now processor x will take 4 threads at once while y will take all 6. Lets assume x has a higher IPC than y. Let each thread take 2 seconds for execution per core in cpu x while cpu y takes 6 seconds to process the same thread per core.

Cpu x will take two cycles to execute 6 threads as it has maximum 4 cores. So the total will be 4 secs to execute 6 threads.
Cpu y will execute all 6 threads in parallel and complete the task in 6 seconds in a single cycle.

We see that despite cpu x having a lack of two cores processed the same process two seconds faster than cpu y which had an advantage of two cores.
The only advantage that cpu x had is single threaded performance or simply cpu time required to execute a single thread of m instructions.

I hope you get the point now mate.:smile:
Click to expand...

Nice explanation Vicky. It clearly shows even if a processor does have less number cores than another but has far superior single-threaded performance, it can actually beat the one with more number of cores. Rep + for you.
 
$$Lionking$$

$$Lionking$$

In the zone
Listen mate more cores are good as they facilitate threading but its not that simple. IPC also has to be higher as well as efficiency.

Vicky - Higher IPC is good. But the performance is NOT dependant only on IPC, but on many other factors. Cores matter as much as IPC do and probably more today with everything moving towards GPGU and all.

Currently intel trounces amd in both these scenarios.

^^If you really think this to be true then you should get your facts set straight. Run wprime 32/1024 on an Intel quad core and an AMD 6 Core and you're in for a surprise my friend. AMD X6 will beat the living crap out of Intel Quad Core inn multi threaded apps.

From the link in the end here is something for you -

wprime 1024m

Intel Core i7 2600k - 245.201 sec
Phenom II X6 1090T - 260.458 sec
Phenom II X6 1055T - 297.836 sec
Phenom II X4 980BE - 334.54 sec
Core i5 2500K - 352.733 sec
Core i5 2400 - 375.476 sec


AMD X6 1055T beats Intel i5 2400 or for that matter even 2500k, So does the 1090T(Its right up there with i7 2600k which costs about twice as much.)
Core i5 2400 is THE VERY LAST IN THE RACE IN A MULTI-THREADED SCENARIO.

I'm damn sure that SQL server (heck even the mighty oracle server) will fall in front of sandybridge. Try it to believe it.

Have you ever tested it?? Database applications are highly multithreaded. Your mighty Oracle is licensed per core. They have a reason for that. Otherwise they would be licensing per some other performance measure and AMD would have been obsolete from database centric server, which it isnt.


All the tasks you mentioned are a piece of cake for the i5 2400 and yes its faster than amd's 6 core offerings in all these tasks no matter how heavily threaded they are.

X6 1055T will anyday beat i5 2400 in any database related application. Combine that with the high amount of threaded workload the OP is going to put on the machine i5 will not stand a chance against X6(Im even surprised that he posted his current machine gets - slow as hell - when it should just have just STOPPED!!!).

Your 2 second 4 core/ 6 second 6 core math is also wrong unfortunately - Intel does not have . Its more like 4sec 4core / 4.8 sec 6 Core (AMD being 20% behind in IPC).

So when you calculate 6 threads its 8secs for Intel & 4.8secs for AMD.

Amd's current scenario is highly skeptical and is not that future proof after bulldozer's dismal performance.

Current scenario is sceptical - I totally agree! I HATE RORY READ! He's an epic A**. I HATE WHAT HE DID TO AMD. That is NOT the AMD way! Jerry Sanders' policy was - "People first, products and profits will follow!"

and 8150's performance is not dismal - its disappointing - we were all expecting better right??

8150 performs between 2500K & 2600K and is priced accordingly in between them. Unless you think performance of 2500K or 2600K is dismal I dont know how 8150's performance is dismal!!????? IPC sucks but things will get sorted out in probably a years time. And I dont think the OP will upgrade cpu before 1 year atleast??!!

On the other hand, intel's ivybridge offereing for 1155 socket is much much more promising and offers a great upgrade path as it will fit all existing h67,h61,p67 and z68 boards.

Yeah I have not seen any official announcement from Intel and unless that happens Im not believing you. There has been some news that pins might be compatible but u still hv to buy a new board etc.. ALL RUMOURS!!!!!!

What the fact is that Intel has changed 3 sockets in 4 years!

See this as well:

**www.hardwarecanucks.com/forum/hardware-canucks-reviews/44339-intel-core-i3-2120-core-i5-2400-lga1155-processors-review-6.html

In conclusion - i5 2400 for the money - a bad choice vs 1090T!!! Enjoy! ;)

If he were overclocking > Core i5 2500K was the way to go eyes closed - very awesome OC potential and performs really good at higher clocks and is only about 10k!! :)
 
Cilus

Cilus

laborare est orare
First of all Vicky was not comparing 1090T with 2400 or 2500, he was comaring 1055T and there is no doubt that Sandybridge offering beats it in most cases, especillaly in gaming, which a normal user will do.
I really doubt that here OP is planning to use huge database operations where he can get the benefit from a 6 core CPU.
2ndly FX 8150 can only compete with 2500K in very highly muti-threaded apps, not any other area. Check ou the HardOCP review where in gaming performance it was crushed by 2500K. You have to look into the power consumption too.

I really like your idea of getting a Phenom II quad core for OP's type of use but if he wants future proof solution, i5 2400/i5 2500 is better choice than 1055T.

And 1090T is bearely keeping with 2400 with only heavily multithreaded scenarios, not anywhere else. Even the highly multi-threaded Adobe suits are also performing better in 2400. In gaming and other works it is simply beaten by the i5 2400, if you beleive the Anandtech CPU bench.
 
skeletor

skeletor

Chosen of the Omnissiah
Surely using MS SQL Server 2008 R2?

Prefer Phenom II X6 core in that case. :)

With Piledriver coming on AM3+, I'd say both AMD and Intel platforms are going to see one generation of processors more. Nothing is "future proof" except the cabinet and PSU.

Real world performance difference i.e. in basic tasks isn't noticeable in current gen processors. All are more than fast enough. Just decide upon your work load and then make a choice accordingly. Quite frankly, rather than being Phenom II X6 vs i5-2400, it should be going down to the motherboard imho.
 
Status
Not open for further replies.
Top Bottom